Re: Testing transitions before uploading to unstable
Lars Wirzenius writes:
> Er, no. The point is to get things tested and hopefully fixed before
> uploading to unstable. Experimental might work, but then packages there
> need to be tested with the existing unstable in ways they are not, as
> far as I know, currently being tested.
I disagree. The point of unstable ought to be to test and hopefully fix
things before they propagate to testing. U*un*u seems capable of making
progress here; we seem more interested in non-developers running
unstable.
However, I think your point that developers need to be more diligent in
testing things is entirely correct. The fact that we still need users to
test unstable means that we are not picking up the slack. People ought
to become maintainers because they want to do that, not because they
want to add their favorite software.
This is only IMHO, of course. I know I have been bad at it, and there is
certianly a lot of inertia associated with current expectations of what
we are providing. I haven't been in #debian for a while, but the fact
that there is *ever* a "F*CKED: unstable" /topic in there (if they also
use that kind of language) is absurd.
To reiterate, I don't think there should be less testing of unstable. I
think there should be *more* testing of unstable by the people who are
motivated and qualified. We all volunteered for this job, right?
(I had no problems with using xorg from experimental that were not
trivial and already being fixed, FWIW. And compiling from an unstable
chroot/using piuparts should not be difficult for anyone.)
(I also have two computers sitting next to me; the one that needs to
actually stay up and running uses etch. You may probably be running
stable on something, even.)
--
things change.
decklin@red-bean.com
Reply to: